A team from Mother and Infant Research Activities (MIRA) in Kathmandu, Nepal undertook a low-cost, community-based participatory intervention with women's groups to test the impact of interpersonal communication on birth outcomes in an economically poor rural population. MIRA indicates that the estimated infant mortality rate in Nepal is 64 per 1000 livebirths, the neonatal mortality rate 39 per 1000 livebirths, and the perinatal mortality rate 47 per 1000 births. In rural areas, 94% of babies are born at home, and only 13% of births are attended by trained health workers. In the mountainous and rural Makwanpur district, the research site, the district hospital has facilities for antenatal care and delivery, but there are 7852 people per 1 hospital bed.

Researchers first pair-matched 42 geopolitical clusters, selected 12 pairs randomly, and randomly assigned one of each pair to intervention or control. A team of local enumerators then mapped the 24 village development committees on foot, identified and allocated a unique identification number to every household, did a baseline census of demographic and socioeconomic indicators, and generated a list of female household members that was then converted into a list of women meeting the inclusion criteria for the cohort (consent given; age 15-49 years; married; and potential to become pregnant). A team of 44 field interviewers visited every potential member of the cohort.

Surveillance began in February 2001, and involved 28,931 participants in 28,376 households. The local female enumerator visited all cohort members in the ward she was responsible for every month over the study period to record menstrual status. The cluster interviewer conducted 2 interviews for every pregnancy: the first at 7 months of gestation and the second at 1 month postpartum; the interview, which was modular to deal with different outcomes, was developed in Nepali and piloted and repiloted by the local team.

In each intervention cluster (average population 7000), a facilitator - a literate locally resident woman - convened 9 women's group meetings every month. The activities of one facilitator rapidly reached a high proportion of pregnant women, even in poor and remote communities. Only 8% of married women of reproductive age ever attended a group, but the groups attracted 37% of newly pregnant women, and members raised awareness of perinatal issues outside the groups themselves.

"One result of the process was that women sought more information about perinatal health."

In general, healthier behaviours were observed in intervention clusters, Women in these clusters were more likely than those in the control clusters to have had antenatal care, to have taken haematinic supplements, to have given birth in a health facility, with a trained attendant or a government health worker, to have used a clean home delivery kit or a boiled blade to cut the umbilical cord, and to have been cared for by a birth attendant who washed her hands. No differences were noted in delayed wrapping of newborn infants, early bathing, or breastfeeding. Rates of maternal morbidity were similar, but women in intervention clusters were more likely than those in control clusters to have visited a health facility in the event of illness. Likewise, infant illness was more likely to have led to a visit to a health facility.

In 11 cluster pairs, neonatal mortality rates were lower in the intervention group. The pooled rate in the intervention group was nearly 30% lower than in the control group study. The maternal mortality ratio was about 80% lower with intervention than with control clusters. Stillbirth rates did not differ between intervention and control clusters.